Kv4.2 channel activity controls intrinsic firing dynamics of arcuate kisspeptin neurons

Key points Neurons in the hypothalamus of the brain which secrete the peptide kisspeptin are important regulators of reproduction, and normal reproductive development. Electrical activity, in the form of action potentials, or spikes, leads to secretion of peptides and neurotransmitters, influencing the activity of downstream neurons; in kisspeptin neurons, this activity is highly irregular, but the mechanism of this is not known. In this study, we show that irregularity depends on the presence of a particular type of potassium ion channel in the membrane, which opens transiently in response to electrical excitation. The results contribute to understanding how kisspeptin neurons generate and time their membrane potential spikes, and how reliable this process is. Improved understanding of the activity of kisspeptin neurons, and how it shapes their secretion of peptides, is expected to lead to better treatment for reproductive dysfunction and disorders of reproductive development. Abstract Kisspeptin neurons in the hypothalamus are critically involved in reproductive function, via their effect on GnRH neuron activity and consequent gonadotropin release. Kisspeptin neurons show an intrinsic irregularity of firing, but the mechanism of this remains unclear. To address this, we carried out targeted whole‐cell patch‐clamp recordings of kisspeptin neurons in the arcuate nucleus (Kiss1Arc), in brain slices isolated from adult male Kiss‐Cre:tdTomato mice. Cells fired irregularly in response to constant current stimuli, with a wide range of spike time variability, and prominent subthreshold voltage fluctuations. In voltage clamp, both a persistent sodium (NaP) current and a fast transient (A‐type) potassium current were apparent, activating at potentials just below the threshold for spiking. These currents have also previously been described in irregular‐spiking cortical interneurons, in which the A‐type current, mediated by Kv4 channels, interacts with NaP current to generate complex dynamics of the membrane potential, and irregular firing. In Kiss1Arc neurons, A‐type current was blocked by phrixotoxin, a specific blocker of Kv4.2/4.3 channels, and consistent expression of Kv4.2 transcripts was detected by single‐cell RT‐PCR. In addition, firing irregularity was correlated to the density of A‐type current in the membrane. Using conductance injection, we demonstrated that adding Kv4‐like potassium conductance (gKv4) to a cell produces a striking increase in firing irregularity, and excitability is reduced, while subtracting gKv4 has the opposite effects. Thus, we propose that Kv4 interacting dynamically with NaP is a key determinant of the irregular firing behaviour of Kiss1Arc neurons, shaping their physiological function in gonadotropin release.


Introduction
Hypothalamic kisspeptin-GPR54 signalling is critical in regulating reproductive functions in mammals (d' Anglemont de Tassigny & Colledge, 2010). Disruption of this pathway leads to infertility in both sexes, as it prevents maturation of the gonads during puberty (de Roux et al. 2003;Seminara et al. 2003;Kirilov et al. 2013). The neuropeptide kisspeptin is produced in two distinct regions in the rodent hypothalamus: one localised in the rostral periventricular area of the third ventricle (Kiss1 RP3V ) and another in the arcuate nucleus (Kiss1 Arc ). The release of kisspeptin from these neurons is able to activate the G-protein coupled receptor KISS1R (also known as GPR54), expressed by gonadotropin-releasing hormone (GnRH) neurons (Han et al. 2005;Piet et al. 2014), which in turn stimulates GnRH neuron firing and causes the secretion of GnRH into the hypophyseal-portal blood system.
Although both groups of kisspeptin neurons appear to synapse directly onto GnRH neurons, they show distinct properties. Notably, Kiss1 RP3V neurons synapse onto GnRH neuron cell bodies and proximal dendrites, and exhibit a sexual dimorphism, where the females have a larger neuronal population (Clarkson & Herbison 2006;Kauffman et al. 2007). In contrast, Kiss1 Arc neurons, whose population is relatively similar in size in both sexes (Clarkson & Herbison, 2006;Hoong Yip et al. 2015), synapse onto the more distal projections of the GnRH neurons within the median eminence region.
It is not entirely clear how these two populations regulate GnRH neuron firing. Recently, optogenetic stimulation was used to investigate how these kisspeptin neurons might integrate their inputs, and to test the firing conditions required for release of luteinizing hormone (LH) into the bloodstream. Firstly, optogenetic activation of GnRH neurons revealed that these cells require stimulation for at least 2 min at 10 Hz to evoke an LH surge (Campos & Herbison, 2014). Secondly, the excitation of GnRH neurons by upstream neurons was addressed (Han et al. 2015), revealing that Kiss1 Arc neurons also need to fire for 2-5 min at 10 Hz in order to increase LH in the blood to physiologically-effective levels, via a kisspeptin/GPR54 mediated mechanism. Thirdly, the synaptic pathway through which Kiss1 Arc neurons recruit other kisspeptin neurons has also recently been uncovered (Qiu et al. 2016). Briefly, Kiss1 Arc neurons were shown to display a frequency-dependent differential release of peptides (kisspeptin, neurokinin-B and dynorphin) and amino acid neurotransmitters (glutamate), which, at high firing frequency (20 Hz), evoked synchronised activity of Kiss1 RP3V neurons, as well as ipsilateral and contralateral Kiss1 Arc neurons.
Although the circuitry in which kisspeptin neurons are embedded is slowly being unravelled, characterisation of the intrinsic electrophysiological properties of these cells remains incomplete. Several studies have shown that a variety of ion channels are present in these cells Zhang et al. 2013;Piet et al. 2014), but it is still unclear how the distribution and interaction of ion channels determines their spiking patterns and overall excitability.
In this paper, we have used a Kiss-Cre:tdTomato transgenic mouse (Yeo et al. 2016) to further characterise the electrical properties of Kiss1 Arc neurons in males. Using patch clamp recording combined with conductance injection, we show that these cells have firing properties strongly influenced by fast inactivating potassium currents (A-type current). This current imposes sub-threshold membrane oscillations, which by interacting with other subthreshold currents -like persistent sodium currents (NaP) -generates spiking irregularity, even in the absence of synaptic activity. By characterising the A-type current through means of pharmacological blockade and kinetics, combined with single-cell RT-PCR, it is demonstrated that this current is mediated mainly by Kv4.2. Using conductance injection, we were able to effectively control spiking irregularity and excitability in Kiss1 Arc neurons by modulating the level of Kv4-type conductance. These results identify an important determinant of the firing dynamics of Kiss1 Arc neurons. The heterogeneous levels of Kv4 conductance present in these cells not only diversify their spiking patterns, but also control their excitability, suggesting that it might be a strong candidate for modulation in order to regulate Kiss1 Arc excitation onto GnRH neurons and other kisspeptin neurons.

Ethical approval
Animals were killed in accordance with the UK Home Office regulations under the Animal (Scientific Procedures) Act of 1986. The establishment, breeding and care of the mouse line were approved by a Local Ethics Committee at the University of Cambridge and performed under authority of a Home Office Licence (UK). 2016). Briefly, Cre recombinase was inserted in exon 2 after the first five amino-acids of the KISS1 protein sequence (Met-Ile-Ser-Met-Ala). Next, the line was crossed with a reporter line expressing tdTomato under the constitutive CAG promoter, but preceded by a floxed STOP codon (Jackson Laboratories, Bar Harbor, ME, USA). The resulting offspring selectively expressed the tdTomato reporter only in Kiss1-expressing cells. Since the Cre recombinase gene disrupts the Kiss1 coding sequence, Kiss-Cre homozygous mutant mice had impaired gonad development and thus were infertile. Heterozygous animals, which were all genotyped, show normal fertility and were used for breeding and for experiments.
All data shown are from heterozygous mice, with the exception of the cells used for neurobiotin-filled images (Fig. 1A), for which we used homozygous mutant mice, as strong tdTomato expression in fibres allowed clearer confocal images. In order to exclude the physiological effects of the fluctuating levels of oestrogen throughout the oestrous cycle, only adult males were used for experiments.

Patch-clamp recordings
Electrophysiological recordings were carried out with an Axon Multiclamp 700B amplifier and CV-7B headstage (Molecular Devices, Sunnyvale, CA, USA). Custom written Matlab software (Mathworks, Natick, MA, USA) was used to control a 16-bit National Instruments X-series board, with sampling and waveform generation at 20 KHz. Inputs were filtered before sampling using the built-in 4 KHz Bessel filtering of the Multiclamp. The Multiclamp Commander software was used to obtain a bridge balance (current-clamp experiments) and series resistance compensation (voltage-clamp experiments; <30 M and compensated up to 90%). Borosilicate glass capillaries (5-7 M , Harvard Apparatus, Kent, UK) were used to pull patch pipettes with a PP-83 puller (Narishige, Tokyo, Japan), which were then fire-polished (Micro Forge MF-830, Narishige) to promote high-quality seal formation.
Passive properties were determined by applying a hyperpolarising step current (from −20 to −40 pA) in current-clamp mode, and fitting an exponential curve to the response: where, R is the input resistance, I is the step current injected, τ is the membrane time constant and V rest is the resting membrane potential.
Firing properties were assessed in current-clamp configuration, where synaptic activity was reduced by 10 μM gabazine, 10 μM CNQX and 10 μM APV. Firing irregularity was quantified by applying a square current step (2-10 s) sufficient to evoke firing at 7-13 Hz. Next, the interspike intervals (ISI) were used to calculate the coefficient of variation of the interspike interval (CVISI), defined as the ratio of the standard deviation of the ISI to the mean ISI.

Artificial conductance injection
Kv4-type conductance was artificially modulated in Kiss1 Arc neurons using the dynamic-clamp technique (Robinson & Kawai, 1993;Sharp et al. 1993a;Robinson, 2013). In summary, a hard real-time SM2 system (Cambridge Conductance; see Robinson, 2008) was used to sample the membrane potential and inject a fast-inactivating outward conductance according to the Hodgkin-Huxley-type equation: whereḡ is the maximum conductance; V m is the membrane potential; E K is the K + reversal potential; and Right: detailed morphology of a neurobiotin-filled neuron. Scale bars of 150 μm and 50 μm respectively. B, around 22% of the recorded cells displayed spontaneous firing, even in the presence of synaptic blockers. C, when stimulated with a square step current, most cells displayed a clear irregular-spiking pattern, noticeably at lower firing frequencies. D, in other neurons, spiking irregularity was also present at higher firing frequencies. E, distribution of 5104 interspike intervals from a cell firing at 7-13 Hz. The broad distribution shows no distinct periodic peaks. Kurtosis: 8.28 Skewness: 1.60. F, kisspeptin neurons resisted firing at frequencies higher than approximately 40 Hz, as strong excitation led to depolarisation block. G, hyperpolarisation induced current (H-current) observed in a fraction of kisspeptin neurons (14/49; seen here as a depolarising 'sag'). m and h are the activation and inactivation variables, respectively, which obey: where m Ý and h Ý are the voltage-dependent, steady-state limiting values of m and h; and τ m and τ h are the voltage-dependent time constant of activation and inactivation, respectively. The voltage-dependent quantities m Ý , h Ý , τ m and τ h were specified as follows: and The Kv4 conductance kinetics were determined from currents previously recorded in mouse cortical interneurons (Mendonça et al. 2016), which displayed virtually identical kinetics to those observed in Kiss1 Arc neurons. To facilitate interpretation of dynamic-clamp experiments, the amount of artificial Kv4-type conductance (g Kv4 ) injected is characterised by its peak value at 0 mV (g max , 0 mV).

Confocal imaging
In some cases, neurobiotin (0.5%) was added to the intracellular solution in order to analyse the morphology of Kiss1 Arc neurons. Cells were patched for at least 20 min in order to allow complete diffusion of neurobiotin into finer processes of the neuron. Next, the slices were fixed in 4% paraformaldehyde overnight, followed by 3 × 15 min phosphate-buffered saline (PBS) washes. Alexa 405 Fluor Streptavidin conjugated dye (1-2 μg mL −1 , Life Technologies) was then applied to the slice for 3 h at 5°C. Another series of PBS washes was done before mounting the samples on microscope slides with ProLong Gold antifade reagent (Invitrogen), which were then kept in the dark at 4°C for at least one day before imaging. Confocal images were acquired with a Leica SP2 microscope, and a minimum 2 μm resolution z-stack was acquired for each cell.

Cytoplasmic harvesting and single cell reverse-transcription PCR
Single cell cytoplasm collection was carried out using a similar approach to that described previously (Yano et al. 2006;Subkhankulova et al. 2010). In summary, using an intracellular solution containing an RNAse inhibitor (RNase OUT, 1 U μL −1 , Life Technologies), a standard whole-cell recording mode was established. Next, a gentle negative pressure was applied (<150 mbar), creating a slow inward flow of the cytoplasm into the patch pipette. Proper care was taken to maintain the seal resistance. After approximately 2 min, most of the cytoplasmic content was collected, whereupon a slightly stronger pressure was applied in order to fully aspirate the nucleus. The contents of the pipette were then expelled into PCR tubes containing random primers, oligo-dT primers, dNTPs (Promega, Southampton, UK), RNaseOUT (Life Technologies, Paisley, UK) and sterile MQ water for cDNA synthesis via reverse transcription.
Briefly, the presence of Kv4 channel mRNA in kisspeptin neurons was verified with single cell nested PCR reactions, in particular to detect Kv4.1, Kv4.2 and Kv4.3 gene expression. The cDNA was amplified with two consecutive PCRs, consisting of 40 cycles on the first round, 20 cycles on the second, and a 1:100 dilution of the product between first and second rounds, using two sets of primers (nested primers) for each gene of interest (See Table 1). The resulting product was analysed for appropriate size using 1.5% agarose gel electrophoresis. The Hprt housekeeping gene was used as an internal positive control to ensure that all cDNA conversions were successful. The pro-opiomelanocortin (POMC) gene was used as an internal negative control to confirm that all tdTomato cells, and only tdTomato cells, expressed Kiss1 mRNA. Additionally, sterile distilled water replacing cDNA was also used as a negative control.

Data analysis and statistics
Data were analysed with custom Matlab scripts. All statistical values described in this work are shown as mean ± standard error of the mean (SEM), unless otherwise stated. In most experiments, adequate demonstration of normality of distribution was not possible, and so non-parametric tests, like the Mann-Whitney U test, were applied, except where stated.

Kiss1 Arc neurons display an irregular firing pattern
Kisspeptin-expressing neurons were distributed over a large portion of the arcuate nucleus, notably in the caudal region. These cells had very small somas, with J Physiol 596.5 long dendritic projections that were mainly oriented dorsally (Fig. 1A), with a fairly high input resistance of 1.34 ± 0.69 G , passive membrane time constant of 41.9 ± 18.55 ms, capacitance of 33.7 ± 13.68 pF and resting membrane potential of −67.4 ± 8.17 mV (mean ± SD, n = 49). Approximately 22% of cells (11/49) displayed spontaneous spikes at around 1 Hz (Fig. 1B). Interestingly, when a step current injection was applied (from 2 to 10 s), most cells presented an irregular spiking pattern with sub-threshold fluctuations, which was usually, but not always, more pronounced at lower firing frequencies ( Fig. 1C-E). Kiss1 Arc neurons resisted firing at high frequencies ( Fig. 1F), reaching a maximum of approximately 40 Hz, and exhibiting depolarisation block with strong current stimulation. Additionally, 29% (14/49) of the cells displayed a noticeable H current (Fig. 1G) as described by others in Kiss1 neurons (Gottsch et al. 2011;Zhang et al. 2013). Although these different features illustrated the variation in electrophysiological properties of Kiss1 Arc neurons, no clear pattern emerged sufficient to suggest the existence of distinct populations.

Kiss1 Arc neurons display sub-threshold membrane fluctuations and persistent sodium currents
Irregular-spiking kisspeptin neurons displayed highly variable interspike intervals (ISIs), characterised by noisy membrane potential fluctuations at the firing threshold, especially during longer intervals ( Fig. 2A-C), and with a noticeable low frequency component (up to 30 Hz, Fig. 2D). As noisy sub-threshold oscillations may perturb spike timing precision, they were assumed to be an important factor contributing to the generation of irregular firing. These fluctuations persisted throughout ISIs, with no apparent trend in amplitude over time during ISIs, suggesting that they were likely produced by non-inactivating voltage-dependent currents. Persistent sodium currents (NaP), which have already been described in Kiss1 RP3V neurons , and are a likely candidate for involvement in the depolarizing phase of oscillations. We therefore sought to measure the level of NaP in Kiss1 Arc neurons.
In the presence of potassium and calcium channel blockers (TEA 2 mM, 4-AP 2 mM and Cd 2+ 200 μM), Kiss1 Arc neurons were subjected to a slowly depolarising voltage ramp (20 mV s −1 ), revealing a TTX-sensitive inward current, activating at around −60 to −55 mV (Fig. 2E, n = 9 total, n = 4 with TTX application), and with no apparent inactivation, characteristic of NaP. Based on this voltage range of activation it can be assumed that NaP is involved in the generation of sub-threshold fluctuations. However, NaP is expressed in numerous types of neurons (Bean, 2007), and is not necessarily related to firing irregularity. We therefore looked for other sub-threshold currents that could promote ISI noise, particularly currents that could interact with NaP.

An A-type current mediated by Kv4 is highly expressed in Kiss1 Arc
We attempted to characterise outward currents that were expressed in Kiss1 Arc neurons. When depolarising voltage steps were applied, these cells displayed a prominent A-type K + current, with a fast activating (3 ms time to peak) and fast inactivating (30 ms) component. Surprisingly, this was the dominant outward current type expressed in Kiss1 Arc (Fig. 3A, n = 13). A-type current is categorised as a sub-threshold current (Bean, 2007), and since it was present in virtually all viable cells patched, we investigated its properties in more detail.
We used pre-pulse steps (Burdakov & Ashcroft, 2002;Amarillo et al. 2008;Maffie et al. 2013) to isolate the A-type current by subtraction, and measure its recovery from inactivation (Fig. 3B, n = 17), as well as steady-state activation and inactivation relationships (Fig. 3C). Using the pre-pulse method in the presence of 5 mM TEA also allowed accurate isolation of the A-type current from other outward currents expressed in Kiss1 Arc neurons . E, responses to slow depolarizing membrane potential ramps show salient non-inactivating (persistent) sodium currents activating over the voltage range spanned by sub-threshold oscillations. Inward peak current was 27.9 ± 2.6 pA at −50 mV, using a linear leak subtraction.

. Kisspeptin neurons display a prominent A-type current
A, voltage steps from −50 to 50 mV reveal a fast-inactivating outward current activating at around the spike threshold potential. B, illustration of the series of inactivating voltage steps that were used to characterise the kinetics of the A-type current. C and D, data from A and B were used to reconstruct the steady-state activation and inactivation curves (C), as well as the time constant for recovery of inactivation (D). E, inactivating pre-pulses (inset) were applied in order to isolate the fast inactivating current component (in the presence of 5 mM TEA), which allowed characterisation of its activation and inactivation dynamics. For each voltage step, the isolated A-type current was obtained by subtracting the control step (preceded by an inactivating step) from the test step. Kisspeptin neurons showed 16.2 ± 4.9 nS of peak conductance at 0 mV. F, the time constants for activation and inactivation were weakly voltage dependent. G, left: the peak current amplitude is dominated by fast-inactivating currents, particularly at more negative membrane potentials. Right panel: zoomed region of I-V plot shown on the left, leak currents subtracted.
( Fig. 3E-G, n = 16). We found that these cells expressed a total of 16.2 ± 4.9 nS peak conductance at 0 mV. The current activated at around −55 mV, and had a time constant for the exponential recovery of inactivation in the range 15-60 ms (weakly voltage-dependent; Fig. 3F). The peak outward current was dominated by the fast-inactivating fraction, which comprised 70% of the total amplitude at 0 mV (1136 pA for A-type current, and 496 pA for the remaining TEA-resistant current, Fig. 3G). Kv4-type channels are known to contribute substantially to the macroscopic A-type current in many cell types, but in some cases, Kv1.4, Kv3.3 or Kv3.4 may also be involved, as they display similar fast-inactivating properties. However, the kinetics of the A-type current recorded in Kiss1 Arc neurons strongly suggest that it is mainly mediated by somatic Kv4 channels. Specifically, the similar activation and inactivation curves (Birnbaum et al. 2004), the fairly fast recovery from inactivation and the weak voltage dependence of its time course are all known properties of Kv4 channels. Participation of Kv1.4 can be discounted, as it has a time constant for the recovery from inactivation in the range of seconds (Wickenden et al. 1999).

Kv4 transcript expression in kisspeptin neurons at the single-cell level
To corroborate the electrophysiological data, expression of the Kv4 potassium channel subunits Kv4.1, 4.2 and 4.3 was examined by RT-PCR following harvesting of cytoplasm from single Kiss1 Arc cells (Fig. 4E). A total of 15 arcuate tdTomato/kisspeptin cells were screened for Kiss1, Hprt and Kv4 subtype transcripts, and we observed that Kv4.2 expression predominated: 86% of Kiss1 Arc neurons were positive for this Kv4 subtype. In contrast, only 9% of the cells expressed Kv4.3, while no Kv4.1 expression was found. The predominant expression of Kv4.2 suggested that the prominent A-type current was mainly generated by this channel subtype.

Spiking irregularity is correlated with the presence of Kv4 conductance in Kiss1 Arc neurons
Kiss1 Arc neurons displayed an abundant Kv4-type conductance (g Kv4 ) , which was presumed to affect their firing properties, as it activated at fairly low membrane potentials, and would thus be expected to influence the trajectory of membrane potential as it approaches threshold. We therefore anticipated that Kv4 current could affect the spiking irregularity observed in Kiss1 Arc cells, in analogy to its role in irregular-spiking cortical interneurons, as shown previously (Mendonça et al. 2016).
In order to test this possibility, we first quantified the irregularity of kisspeptin neuron firing by estimating the coefficient of the interspike interval (CVISI) at 10 Hz (CVISI 10 , from 7 to 13 Hz). We observed that these cells were not homogeneous in this regard. Instead, they displayed a broad distribution in their spiking irregularity (Fig. 5A, n = 47). While some cells fired with a fairly regular pattern (CVISI 10 < 0.2), most cells displayed an irregular-firing pattern, ranging from relatively mild (0.3) to very high (>0.7) spiking irregularity.
The broad distribution of spiking irregularity found in Kiss1 Arc neurons suggests that these cells might naturally express ion channels in different proportions, resulting in different firing patterns. We were particularly interested to test whether heterogeneity of spiking irregularity was related to distinct Kv4 expression in each neuron. Therefore, in some cells, we assessed both CVISI 10 and pre-pulse isolated g Kv4 , which was possible due to their relative small soma size (space-clamp without 'escaping' sodium currents). When g Kv4 was normalised by the capacitance of each cell (proportional to their membrane area), we found a clear correlation between firing irregularity and g Kv4 density ( Fig. 5B-D, compare C  and D). These findings suggest that the firing variability of Kiss1 Arc neurons could be determined by g Kv4 .

Artificial injection of Kv4 conductance modulates spiking irregularity and excitability
As g Kv4 was apparently associated with the intrinsic spiking irregularity of some Kiss1 Arc neurons, the dynamic-clamp conductance injection technique was used to investigate this link further. This method allows injection of any mathematically-described voltage-dependent conductance in a whole-cell current-clamp recording (Robinson & Kawai, 1993;Sharp et al. 1993b;Robinson, 2013), J Physiol 596.5 allowing us to add or subtract controlled levels of g Kv4 conductance, and test its effect on firing irregularity.
When g Kv4 was altered in Kiss1 Arc neurons, we saw a striking effect on their intrinsic spiking irregularity. When positive g Kv4 was injected in mildly irregular-firing cells, we observed a clear increase in interspike variability, which was normally associated with the presence of prominent noisy sub-threshold 'plateaus' (Fig. 6A and C). In contrast, negative g Kv4 injection in very irregular-firing cells resulted in more regularly distributed action potentials ( Fig. 6B and  C). A clear correlation was found between CVISI 10 and the normalised level of g Kv4 injected (Fig. 6E, n = 17).
However, g Kv4 modulation did not only impact how irregularly Kiss1 Arc neurons fired, but also created a distinct effect on spiking adaptation. Some neurons displayed a significant intrinsic decrease in the firing frequency throughout the constant stimulation, which was usually observed in very irregular-spiking cells. When g Kv4 was reduced in these neurons, they not only produced a more regularised spiking pattern, but also fired steadily at higher frequencies at lower levels of stimulus current (Fig. 6F).

Discussion
Here, we have characterised the electrophysiological properties of arcuate nucleus kisspeptin neurons from a novel Kiss1-Cre:tdTomato transgenic mouse line, providing an insight into the biophysical mechanisms governing their action potential generation. Some of the electrophysiological properties shown here for kisspeptin neurons have already been described by others in different mouse strains, such as the high input resistance, the expression of H-current, persistent sodium current, and spontaneous spiking Zhang et al. 2013;Piet et al. 2014). In this study, we find that these cells also express a strong A-type potassium current. Using pharmacological blockers and single cell RT-PCR, we demonstrated that this current is mainly mediated by Kv4-type channels (likely predominantly Kv4.2). Furthermore, we have highlighted the role that this current plays in shaping the firing pattern of Kiss1 Arc neurons. By interacting with subthreshold inward currents (like NaP), these channels can create interspike voltage fluctuations, which affect the timing of spikes and diversify their firing patterns. Using dynamic-clamp conductance injection, we have shown that modulation of g Kv4 not only changed their firing irregularity, but controlled their firing frequency in response to a constant stimulus. These findings are especially relevant when considering that kisspeptin neurons are normally stimulated tonically by neuropeptides like neurokinin-B (NKB; De Croft et al. 2013). Since its early characterisation, A-type current, commonly but not exclusively carried through Kv4 ion channels, has mainly been associated with enabling low-frequency, regular firing activity (Connor & Stevens, 1971). Here, by contrast, we confirm that, under the right conditions, g Kv4 can contribute to sub-threshold fluctuations and cause an increased irregularity in action potential timing. The mechanism we report here is strikingly similar to what we have observed recently  47 cells). B, in some cells with low series resistance (<20 M ) permitting adequate voltage clamp, both spiking irregularity and Kv4 current amplitude were measured. The coefficient of interspike interval (CVISI) at 7-13 Hz firing frequency was then estimated, while the peak Kv4 conductance was normalised to the capacitance of each cell. Pearson correlation P value: 0.019, r value = 0.88, n = 6. C, example segments of firing from a regular-spiking cell (grey) contrasting with an irregular-spiking cell (black). D, corresponding isolated Kv4 conductance waveforms for voltage steps from −90 to −20 mV obtained with the pre-pulse protocol.  Mann-Whitney test, P value = 0.0012 and 6.38 × 10 −5 for positive and negative g Kv4 injection. E, a positive correlation was found between the change in spiking irregularity and the amount of g Kv4 injected (normalised by each cell's capacitance). Pearson's correlation r value = 0.655, P value = 7.8 × 10 −8 . F, in some strongly adapting cells, it was observed that negative g Kv4 injection produced a more regular firing, which could reach higher frequencies with similar or lower current stimulation than control.
in irregular-firing interneurons (Mendonça et al. 2016), where the same current types (A-type and NaP) have been shown to create an almost identical effect on spiking irregularity. However, although we show that Kiss1 Arc neurons also display the same components, we cannot exclude the possibility that other subthreshold currents (e.g. T-type Ca 2+ currents) contribute significantly to this mechanism. Using computational modelling, the dynamical basis of increased irregular firing in cortical interneurons produced by addition of g Kv4 was made clear: both unstable oscillations and slow dynamics in the interspike interval emerge as g Kv4 is increased, through its dynamical interactions with the sodium current, including NaP (Mendonça et al. 2016). It is interesting that cells embedded in completely different physiological contexts, in different brain regions, seem to use a similar control of firing, which may be viewed as a conserved biophysical mechanism for generating intrinsic spiking irregularity for different purposes in the brain.
Even though it is plausible to draw close comparisons between the firing dynamics of irregular-spiking cortical interneurons and Kiss1 Arc neurons, some distinctions need to be made. Most importantly, the currents required to generate spiking irregularity were present in different proportions in these two cell types. While irregular-spiking interneurons displayed a large NaP current (80.9 pA at −50 mV, or 1.51 pA/pF normalised to the overall mean of the capacitance), kisspeptin neurons displayed a lower level (27.9 pA at −50 mV, or 0.83 pA/pF). In contrast, while irregular-spiking cells expressed sufficient Kv4 conductance to fire irregularly (22.3 nS peak at 0 mV, or 0.42 nS/pF), it was even higher in kisspeptin neurons, representing one of the dominant K + conductance fractions (16.2 nS peak at 0 mV, or 0.48 nS/pF). Reflecting the relative dominance of g Kv4 , kisspeptin neurons were more responsive to its modulation. A reduction of g Kv4 translated into a significant decrease in one of the most important outward conductances in these cells, which as expected, markedly increased their excitability, especially in irregular-spiking cells with strong firing adaptation. We propose that the level of expression of functional Kv4 channels, in addition to modulating firing irregularity, acts as a 'gain controller' of kisspeptin neurons, determining the size of responses to their synaptic inputs.
A similar phenomenon, where A-type current acts as a 'gain controller' has been proposed in orexin-expressing neurons in the lateral hypothalamus (Burdakov et al. 2004), where spontaneously firing cells required less hyperpolarising current to reduce firing, when they expressed larger A-type conductances, implying that A-type current raises the effective gain of the firing frequency-stimulus current relationship. Interestingly, in addition to orexin and kisspeptin neurons, there are reports of several other cell types in the hypothalamus displaying A-type currents (Burdakov & Ashcroft, 2002;Zhang & van den Pol, 2015), suggesting that hypothalamic neurons may use A-type conductance modulation (or specifically Kv4 conductance) as a common strategy to control their excitability.
The surprising heterogeneity of spiking irregularity found in kisspeptin neurons, associated with different levels of g Kv4 expression, suggests a natural fine tuning in these cells, which due to temporary or permanent adjustment of Kv4 channel kinetics, might allow each Kiss1 Arc neuron to have distinctive influences on GnRH neurons and/or other kisspeptin neurons. As the frequency and temporal structure of firing can determine the differential release of neurotransmitters (Dutton & Dyball, 1979;Liu et al. 2011;Schöne et al. 2014;Qiu et al. 2016), the level of Kv4 expressed in each cell can favour the release of different neuropeptides or amino acid neurotransmitters.
However, it is still unclear if these cells can actively change their firing properties by altering the total level of somatic g Kv4 . The gating kinetics of Kv4 channels are known to be modulated by accessory proteins (K + channels interacting proteins, KChIPs) and membrane lipids (Birnbaum et al. 2004;Oliver et al. 2004), while some neuromodulators, like endocannabinoids and cholecystokinin, have been shown to modify Kv4-type current amplitude (Burdakov & Ashcroft, 2002;Amorós et al. 2010). We tested the effect of perfusing three different neuromodulators, neurokinin-B (10 μM, n = 7) and dynorphin-A (50 μM, n = 5), known to affect the excitability of kisspeptin neurons, and cholecystokinin (100 pM, n = 3; 40 pM, n = 3; 50 pM, n = 2; 100 nM, n = 1) which has been reported to increase A-type conductance in arcuate cells (Burdakov & Ashcroft, 2002). These peptides produced no noticeable acute effects on g Kv4 in kisspeptin neurons (data not shown), but it is still possible that other neuromodulators are capable of doing so. It is also possible that g Kv4 is not subject to significant neuromodulation in Kiss1 Arc neurons, but that variation in g Kv4 density may be determined by connectivity and particular role of a neuron within the circuit, and by stochastic gene expression and trafficking of the channel protein.
In summary, we propose that the density of Kv4, predominantly Kv4.2, channels in arcuate kisspeptin neurons J Physiol 596.5 is an important factor that regulates their firing pattern and excitation of GnRH neurons and other kisspeptin neurons. When Kv4 is strongly expressed, Kiss1 ARc neurons fire at very low frequency and asynchronously, likely contributing to the release of amino-acid neurotransmitters. However, when Kv4 channels are downregulated or inhibited, these cells are capable of delivering a more reliable stimulation at a higher firing frequency, favouring the release of neuropeptides and effectively recruiting other kisspeptin neurons and GnRH neurons.